France, Denmark, Finland and the Netherlands temporarily took government websites offline on Wednesday after web developers using artificial intelligence identified a bug that researchers estimate could affect up to 70% of the world’s websites.
A co-founder of Google’s Wiz cybersecurity team told The Wall Street Journal the exploit was "the most severe AI vulnerability ... ever found," adding that humans likely would not have discovered it without AI. That assessment was made on the basis of the vulnerability’s scope and how readily it can be identified by machine-assisted analysis.
Japan’s prime minister also urged citizens to "take action tonight" to install newly released fixes intended to mitigate the flaw. Those patches were published publicly to reduce immediate risk and allow system administrators to restore services safely.
However, a cybersecurity researcher speaking to Reuters warned that AI will also enable attackers to reverse-engineer the public fix, sparking a "race to exploit" the vulnerability. The researcher said the next likely step would be the "mass compromise of every vulnerable web server on the internet."
In response to the disclosures, several governments limited services temporarily to reduce exposure and buy time for patch deployment. The incident highlights a dual effect of AI in cybersecurity: it can accelerate the discovery and remediation of serious bugs while also potentially speeding attackers’ ability to weaponize those same flaws.
(Report by Brendan Ruberry.)
